Simple Sentences in Malagasy
Charles Randriamasimanana
《語言暨語言學》專刊外編之五
《百川匯海：李壬癸先生七秩壽慶論文集》
張永利 黃美金 何大安/編輯
中央研究院語言學研究所
LANGUAGE AND LINGUISTICS MONOGRAPH SERIES NUMBER W-5
STREAMS CONVERGING INTO AN OCEAN:
Festschrift in Honor of Professor Paul Jen-kuei Li on His 70th Birthday
Edited by Henry Y. Chang, Lillian M. Huang, and Dah-an Ho
Institute of Linguistics, Academia Sinica, Taipei, Taiwan
2006
Streams Converging Into an Ocean, 71-96
2006-8-005-004-000003-1
Simple Sentences in Malagasy∗
Charles Randriamasimanana
University of Oxford
This paper examines the notion of Control and how it contributes to our
understanding of the make-up of a typical Malagasy sentence in terms of a core as
well as a periphery, two rather distinct zones. Its main purpose is to explore the
importance of this specific feature for the distinction between an argument and an
adjunct. In the process, it will shed some light on what type of constituent is likely
to undergo the process of incorporation, which one is not and how a typical
adjunct can resort to a switch type mechanism and migrate from one zone to the
other.
Key words: Control, argument, adjunct, incorporation
1. Introduction: Malagasy sentence core & periphery
My main purpose in this paper is to explore the crucial importance of the notion of
Control and its contribution to the distinction between an argument and an adjunct in a
simple Malagasy sentence, in addition to the traditional division of a clause into a core
and a periphery. In what follows, it will transpire that in Malagasy the core of an active
voice sentence may typically1 comprise objects, i.e. a direct object (DO) and an indirect
object (IO), as illustrated in section 2; whereas from among potential preposition
∗
1
This paper is offered to Professor Paul Jen-kuei Li of Academia Sinica on the occasion of his
70th birthday. It was made possible partially by ESRC Project RES-000-23-0505 Verb-initial
grammars: A multilingual/parallel perspective. University of Oxford.
This assumption is based on the following type of rather widespread Malagasy data, where an
overt subject is missing from an active voice sentence:
Nanambady tamin’ny 1938.
N- ana(n>m)-(v>b)ady t-amin’ny
1938.
past-ananavady
past-on the 1938
past-root.have spouse
in
1938
‘Married in 1938.’
From Rajaona (1969:13) Takelaka Notsongaina (Selected Texts), a Malagasy literature
anthology used in Malagasy literature classes throughout Madagascar.
Charles Randriamasimanana
phrases, a Time-Oblique as illustrated in (39) appears to be a typical adjunct lying
outside the core and inside the periphery. Between those two distinct zones, lies a fuzzy
area where an Instrument-Oblique, as described in section 3, is an argument of the
governing verb since the latter necessarily contains the feature [+CONTROL]; by
contrast, there seems to be more variability with the case of Location-Oblique, as shown
in section 4, depending among other things on the presence or absence of the feature
[+CONTROL] inside the relevant verb. Bridging the gap between core and periphery is
incorporation, as envisaged in Randriamasimanana (2004a) and described in section 7
and which hinges on the compatibility of the set of inherent features inside a head verb
with the feature [+CONTROL], as illustrated in sections 5, 6 and 7.
2. Objects in Malagasy
As already shown in Randriamasimanana (1986:655-661), there is a clear
distinction in Malagasy between a direct object (DO) and an indirect object (IO). It will
become apparent that at least one type of DO can be ‘fronted’, despite claims to the
contrary found, for instance, in Paul (2000:37, ex. 24a), Paul (2002), Paul & Potsdam
(2004:248, ex. 11a) and Potsdam (2005, ex. 40), apparently based on data initially released in Keenan (1976:268, ex. 55a).
2.1 Distinction between DO and IO
There is little doubt that in Malagasy both a DO and an IO can be ‘promoted’ to
grammatical subject, thus triggering passive morphology on the verb, as already
reported, for example, in Keenan (1976) and Randriamasimanana (1986).What has been
obscured for the last several years is the simple fact that in this language, at least one
type of DO can be ‘fronted’ despite its being governed by a verb which remains in the
active voice and provided that certain conditions are met. In general, this seems to relate
to quantification and the data contradicts the claim made about DO in Paul (2000:37).
On the other hand, ‘fronting’ is absolutely ruled out for an IO.
2.1.1 Direct object
Despite the fact that a DO and an IO are part of the core in Malagasy, there is a
sharp difference in behaviour between the two, as will be illustrated immediately below.
A non-definite type of DO as illustrated in (5) can be ‘fronted’ in this language -contra
Keenan (1976:268, ex. 55a) -- and the governing verb remains in the active voice, as
shown in (3); in addition, this kind of construction brings with it a partitive reading, as
72
Simple Sentences in Malagasy
seen in (4). By contrast, as will be shown in the next section, an IO simply cannot be
‘fronted’ in the same manner, as this invariably renders the sequence irretrievably
ungrammatical.
(1) Namaky (an’) ilay kitay i Paoly.
N-a(n>m)-(v)aky
(an’) ilay
kitay i
Paoly.
2
2
past-pref.an-root.broken (acc.) DX .sg wood D .sg Paul
‘Paul was chopping the (previous mention) wood.’
(2) Novakin’i Paoly ilay kitay.
No-vak(i)-in(a)
i
Paoly ilay
kitay.
Pass.past-stem.be.broken-ina3 D.sg Paul DX.sg wood
‘The (previous mention) wood was chopped by Paul.’
(3) Ny kitay i Paoly no namaky betsaka.
Ny
kitay i
Paoly no
n-a(n>m)-(v)aky
betsaka.
Topic wood D.sg Paul focus past-pref.an-root3.broken a.lot
‘As far as wood is concerned, Paul was chopping a lot of it.’
(4) Ny kitay i Paoly no namaky.
Paoly no
n-a(n>m)-(v)aky.
Ny4 kitay i
Topic wood D.sg
Paul
focus past-pref.an-root.broken
‘As far as wood is concerned, Paul was chopping some.’
2
3
4
There are three types of determiner in Malagasy based on their case-assignment properties: A
first subtype, D.singular here ‘i’ accompanies a proper name requires Ø case-marker for
nominative, but necessitates the particle ‘an’ for the accusative. Det ‘ny’ ‘(all) the’ represents a
second subtype, which goes on common nouns and which remains invariant since it canNOT
be accompanied by particle ‘an’ even in the accusative case. DX.singular ilay ‘the previously
mentioned’ is the third subtype, which accepts case-marker Ø for the nominative, optionally
takes ‘an’ in the accusative and can go with either a proper name or a common noun.
The distinction between a root and a stem is crucial in Malagasy as, in addition to a difference
in meaning, the first does not involve a stress shift while the second does. Thus in (3) vaky ‘be
acci-dentally broken’ is a root and the main stress falls on the first syllable. Compare that with
stem vaki ‘be deliberately broken’ in (2), with stress on the second syllable as it requires a
stress shift because of suffixation with passive voice affix ‘ina’.
Topic ny ‘as for’ is different from Determiner ny ‘all the (members of a given set)’ in that the
latter presupposes that all members of the set in question are properly identified and that all of
them without any exception are being referred to. This is definitely not the case for Topic ny
‘as for’.
73
Charles Randriamasimanana
(5) Namaky kitay betsaka i Paoly.
N-a(n>m)-(v)aky
kitay betsaka
Past-pref.an-root.broken wood a lot
‘Paul was chopping a lot of wood.’
i
Paoly.
D.sg Paul
Note that the DO in (5) is not definite and that the verb in both (3) and (4) remains in
the active voice and that in both instances, the VP has a partitive reading, the kind of
interpretation to be attributed to example (55a) in Keenan (1976): There is very little
doubt that (55a) originally marked as ‘ungrammatical’ is indeed ‘grammatical’, but with
this other meaning. This situation is very different for (1), where the DO is definite and
in (2) this definite DO has been ‘promoted’ to subject, with the correlate that the VP
does not carry any partitive reading at all.
The phenomenon just described was already illustrated in Randriamasimanana
(1986:659, ex. 182) and very briefly outlined in Randriamasimanana (1994:25). The
grammaticality of sequences (3) and (4) along with that of (55a) in Keenan (1976)
contradicts a crucial claim5 made in Paul (2000:37, ex. 24a) as well as in Potsdam
(2005, ex. 40), for example.
(6) Ny
akoho no
n-ividy i Bao.
Topic chicken focus root.buy
Bao
‘It’s the chicken that Bao bought.’
5
74
Here is a piece of authentic Malagasy data from a Malagasy language forum, adapted from the
website www.wanadoo.mg/forums/read.php? Re: Voyage en Inde, Par: lalaiko, Date: 28
septembre 2005 - 10:58, which shows that a non-definite DO can definitely be fronted while
the verb remains in the active voice.
Ny sakafo afaka mifidy: Balinais, japonais, français, indien ...
Ny
sakafo afaka m-i-fidy
--- : Balinais, japonais, français, indien ...
Topic food
can pres-pref.i-root.choose
: Balinese, Japanese, French, Indian...
Literally: ‘As for food, --- can choose: Balinese, Japanese, French, Indian...’
English: ‘As far as food is concerned, (we) can choose: Balinese, Japanese, French, Indian...’
Note the missing subject ‘we’, as predicted by the assumption made in Footnote 1. The verb
‘m-i-fidy’ ‘present-root.choose is in the active voice form; the corresponding Object-to-Subject
passive form will be: Ø-fid(i)-ina ‘passive.present-stem.be.chosen-suffix.ina’, ‘is-beingchosen’ or ‘is-usually-chosen’; the corresponding Oblique-to-Subject passive form is: Øifidian-ana ‘passive.present-stem.ifidian-suffix.ana’, ‘(the-circumstances-under-which-someone/
something)-is-chosen’.
Simple Sentences in Malagasy
Indeed it is true that sentence (6) does not mean ‘It’s the chicken that Bao bought’.
However it is definitely a fully grammatical Malagasy sequence and it has the following
reading: ‘As far as chickens are concerned, Bao bought some’, with the non-definite
object ‘akoko’ ‘chickens’ in the plural in front and the verb ‘n-i-vidy’ ‘pst-pref.iroot.buy in the active voice. The item ‘ny’ is to be glossed not as definite article ‘ny’,
which means something like ‘(all) the (members of a given set without exception)’, but
rather as ‘topic’ ‘ny’ ‘as far as a number of members of a set are concerned’. The same
crucial problem arises in a whole series of papers, to include but not restricted to Paul &
Potsdam (2004) as well as Paul (2002).
2.1.2 Indirect object
An IO, as illustrated in (7), can be ‘promoted’ to subject as in (9), just like its DO
counterpart in (8). However, a non-definite DO can be ‘fronted’, as in (10), but an IO as
shown in (11) simply can NOT, as this yields an ungrammatical sentence.
(7)
(8)
(9)
(10)
(11)
Nanome (an’) ilay boky an’ i Paoly i Jeanne.
N-an-ome
(acc.) Dx.sg book prep
D.sg Paul D.sg Jeanne.
Past-pref.an-root.given
the book belong.to
Paul
Jeanne
‘Jeanne gave the (previous mention) book to Paul.’
Nomen’ i Jeanne an’ i Paoly ilay boky.
N(o)-ome-n(a)
D.sg Jeanne prep
D.sg Paul DX.sg book.
Pass.past-stem.be.given-by
Jeanne belong.to
Paul the
book
‘The (previous mention) book was given by Jeanne to Paul.’
Nomen’ i Jeanne (an’) ilay boky i Paoly.
N(o)-ome-n(a)
D.sg Jeanne (acc) DX.sg book D.sg Paul.
Pass.past-stem.be.given-by
Jeanne
the
book
Paul
‘Paul was given the (previous mention) book by Jeanne.’
Ny boky no nanome an’i Paoly i Jeanne.
Ny
boky no
n-an-ome
an’ i
Paoly i
Jeanne.
Topic book focus past-pref.an-root.given prep D.sg Paul D.sg Jeanne
‘As far as books are concerned, Jeanne gave Paul some.’
*(*An’) i
Paoly no
nanome boky i
Jeanne.
(*Prep) D.sg Paul focus past-give book D.sg Jeanne
‘To Paul, Jeanne gave a book/books.’
There exist specific tests to distinguish between a DO and an IO, as detailed in
Randriamasimanana (1986:655-661). In example (7), a Binary Branching analysis is
75
Charles Randriamasimanana
adopted, where the second occurrence of ‘an’ with a di-transitive verb is analysed as a
preposition, i.e. part of a non-verbal predicate. At any rate, this detail will not affect the
conclusions reached in this paper.
3. Instrument-Oblique
In Malagasy, although it is not strictly speaking part of the initial core of a simple
sentence, an Instrument-Oblique is an argument since the head verb which licenses the
relevant PP necessarily contains the feature [+CONTROL]; without this positively
valued feature, there would simply not be an Instrument-Oblique, in the first place. This,
of course, raises questions about the grammaticality judgment shown in Keenan
(1976:269, ex. 57), and casts doubts on the claim made in Paul (2000:38, ex. 26a &
ex. 26b), repeated in Pearson (2001:142, ex. 138a & ex. 138b). The ‘circumstantial’
passive voice indeed encodes the ‘promotion’ of an argument Oblique to Subject and
the so-called ‘circumstantial’ or ‘relative’ voice form usually has the shape ‘an…ana’ or
‘i…ana’, depending on the relevant stem of the verb.
3.1 Fronted & non-fronted Instrument-PP in Malagasy
In Malagasy, an Instrument-Oblique can be expressed either in an active voice
sequence as a non-fronted preposition phrase (PP) with ‘Ø-amina’ ‘non-perfective-with’
governed by a [+CONTROL] verb, as in (12), or as the grammatical subject of a verb in
the ‘circumstantial’ passive voice, as in (13) or (14); this suggests that an InstrumentOblique is and remains an argument of the verb independently of whether the verb is in
the active or in the passive voice. In addition, of course, the subject can optionally be
‘fronted’. Thus, in (12), the Instrument-Oblique has not yet been ‘fronted’, whereas in
(13) and (14), it has already been fronted. One further difference is that in (13), the
preposition ‘Ø-amin(a)’ ‘non-perfective-with’ has disappeared, but the verb shows
‘circumstantial’ voice morphology with ‘an…ana’ surrounding the verbal stem; in (14),
on the other hand, the preposition ‘Ø-amin(a)’ persists and at the same time, ‘circumstantial’ voice morphology appears on the verb.
For lack of space, consideration of the version of (13) will be left out, where the
determiner ‘ny’ does not appear within the sentence although this is definitely a
possibility. Sequence (15) below is an ‘authentic’ piece of data displaying this very
feature, adapted from an on-line Malagasy website. Likewise, the alternative sequence
to (13) will not be discussed, where the constituent ‘ny famaky’ ‘the axe’ has not been
fronted and which according to a claim found in Paul (2000:37) should not exist, but
which is a perfectly grammatical sentence in Malagasy. Indeed in (15) the derived
76
Simple Sentences in Malagasy
subject encoding an initial Instrument-Oblique has not been fronted; only the ‘circumstantial voice morphology on the verb ‘prefix.a(n>m)-stem.(v)oah-suffix.ana’, i.e.
‘amoahana’ tells us that this constituent used to be an argument Instrument-Oblique in
the active voice counterpart to (15).
(12) Mamaky kitay (betsaka) amina famaky i Paoly.
famaky i
Paoly.
M-a(n>m)-(v)aky
kitay (betsaka) Ø-amina
pres-pref.an-root.broken wood (a.lot)
nonperf-with axe
D.sg Paul
‘Paul is chopping (a lot of) wood with an axe.’
(13) Ny famaky no amakian’i Paoly kitay.
Ø-a(n>m)-(v)aki-an(a)
i
Ny6 famaky no
Det axe
focus pass.pres-an-stem.be.broken-suff.ana.by D.sg
Paoly kitay.
Paul
wood
Either ‘It is the axe that is being used by Paul to chop (some) wood’
or ‘It is the axe that is habitually used by Paul to chop wood.’
(14) Amin’ny famaky no amakian’i Paoly kitay.
Ø-amin’
ny famaky no
Ø-a(n>m)-(v)aki-an(a)
nonperf-with Det axe
foc pass.pres-an-stem.be.broken-ana.by
i
P. kitay.
D.sg P. wood
Either ‘It is with the axe that Paul is chopping wood’
or ‘It is with the axe that habitually Paul is chopping (some) wood.’
Both (13) and (14) involving a ‘fronted’ Instrument-Oblique show that the so-called
‘circumstantial’ passive voice is mandatory in Malagasy. But even if the InstrumentOblique PP was not ‘fronted’, as in (15) adapted from http://www. haisoratra.org/, the
verb must still be in the circumstantial passive voice with the circumfix ‘an...ana’, as is
quite evident around the verbal stem ‘voah(a)’ ‘be.opened’. Therefore, what is
absolutely crucial is the co-occurrence of the ‘promotion’ of an argument of the verb,
here an Instrument-Oblique ‘promoted’ to Subject, i.e. ‘ny Poezia’ ‘(the) Poetry’ and at
the same time the critically important appearance of circumstantial passive morphology
‘an…ana’ surrounding the stem of the verb.
6
Det ny ‘(all) the (members of a given set)’ in this case refers to a set made up of one single
member, who has already been properly identified either previously in the discourse or
contextually.
77
Charles Randriamasimanana
(15) Amoahan’ny mpanoratra ny hevitra amam-pihetseham-po.
A(n>m)-(v)oah(a)-an’
ny
mpanoratra ny
hevitra
an-stem.be.opened-an(a) Det mpanoratra Topic hevitra
used to express by
Det writers
Topic ideas
amam-pihetseham-po ao
anatiny
koa anefa
ny Poezia...
amam-pihetseham-po ao
anati-ny
koa anefa
ny Poezia…
as well as feelings
there inside-of.them also however Det poetry
Literally: ‘Is also used by writers to express ideas as well as feelings inside
them Poetry however.’
English: ‘Poetry is however also used by writers to express ideas as well as
feelings inside them.’
(16) Mamoaka ny hevitra amam-pihetseham-po ao anatiny.
M-a(n>m)-(v)oaka ny
hevitra amam-pihetseham-po ao
Pres-an-root.open Topic ideas as.well.as.feelings
there
anati-ny
koa anefa
ny mpanoratra amin’ ny alàlan’ny Poezia.
inside-of.them also however Det writers
by-way.of
Det poetry
‘(All) Writers express ideas as well as feelings they have inside them in the
form of poetry.’
In sequence (15), we have the original statement whereas sentence (16) shows an
intermediate structure provided in order to facilitate comprehension and comprising an
Instrument-Oblique, ‘amin’ ny alàlan’ny Poezia’ ‘by way of poetry’, with the preposition amin(a) overtly expressed alongside an active voice verb. A comparison shows
that in (15), the preposition ‘Ø-amin(a)’ ‘non-perfective-with’ has apparently simply
been ‘absorbed’ but the verb is in the circumstantial passive voice: the form ‘an…ana’
surrounds the passive verbal stem ‘voah’. This contrasts with the root voaka for the
active voice form of the same verb in sequence (16).
Last but not least, consider the following sequence involving an InstrumentOblique from this excerpt shown in (17) adapted from on-line Midi Madagascar:
(17) Io moa no nitifirany intelo ilay karana tompon’ny tsena.
Io
moa
no
n-i-tifir-an(a)-ny
This indeed focus past-pref.i-stem.be.fired.on.with-suff.ana-by.him
Intelo
ilay
karana tompo-n’ ny
tsena.
three.times DX.sg (p.m.) Indian owner-of Det shop
‘It was this which was used by him to fire three times upon the Indian owner
of the shop.’
78
Simple Sentences in Malagasy
(18) Nitifitra (an’) ilay karana tompon’ny tsena intelo tamin’io moa izy.
N-i-tifitra
(an’) ilay
karana tompo-n’ ny tsena
Past-pref.i-root.fire.on (acc) DX.sg Indian owner-of. the shop
Intelo
t-amin’
io
moa
izy.
three.times Perf-with this indeed s/he
‘He fired three times on the (previous mention) Indian owner of the shop
with this.’
Once again, the Instrument-Oblique, ‘t-amin’io’ ‘perfective-with this’, is overtly
expressed in the active voice sequence (18) with the verb ‘nitifitra’ ‘past-fire.on’, and
crucially the latter contains the positively valued feature [+CONTROL]. This presence
makes the PP an argument of the verb. This Instrument-Oblique PP has been
‘promoted’ to subject in (17), losing its preposition ‘t-amin(a)’ ‘perfective-aspect-with’
in the process and at the same time, as expected, this loss is compensated by
‘circumstantial’ voice morphology on the verbal stem with the circumfix ‘i…ana’
surrounding the verbal stem ‘tifir’. In addition, there appears to be a very strong
preference for ‘fronting’.
3.2 Instrument-Oblique and published literature
There has been an unfortunate confusion in the published literature as to the
argument status of Instrument-Oblique, probably due to a large extent to sentence (19)
reproduced below. Initially, sequence (19) was released in Keenan (1976:269, ex. 57),
which was re-interpreted as a typical Instrument-Oblique construction, and therefore
emulated in (20), with slightly adapted glosses, from Paul (2000:103, ex. 21c), later
repeated in Paul (2001) as well as in Pearson (2001:142, ex. 138a) and recently reanalysed in Law (2005:178, ex. 3). Nevertheless, a typical Instrument-Oblique involving
passive voice was already available in Randriamasimanana (1986:466, ex. 63) and further
explanations were provided in Randriamasimanana (2004b:271-280) along the lines
sketched above in section 3.1.
(19) Amin’ity savony ity no manasa lamba ny vehivavy.
Ø-amin’
ity
savony ity
no
m-an-(s)asa
Nonpast-with DX.sg soap
DX.sg focus pres-an-root.wash
lamba ny vehivavy.
linen
Det woman
Intended reading: ‘It is with this soap that Rasoa is washing clothes.’
Actual reading: ‘Hey, you women, with THIS soap, go and wash linen!’
79
Charles Randriamasimanana
(20) Amin’ny penina no manoratra aho.
Ø-amin(a)
ny penina no
m-an-(s)oratra
aho.
Nonperf-with Det pen
focus pres-pref.an-root.write I
Intended reading: ‘It is with the pen that I write.’
Sequence (20) where the Instrument-Oblique has been ‘fronted’ and where the verb is
still in the active voice is simply not Malagasy: this preposition phrase is an argument
of the verb manoratra ‘to write’, as the latter clearly describes a DELIBERATE kind of
ACTIVITY; in other words, this verb does quite clearly contain the inherent feature
[+CONTROL], as argued for in Randriamasimanana (1999:522-524) and the preposition
phrase ‘Ø-amin’ny penina’ ‘non-perfective-with the pen’ is certainly NOT an adjunct,
as claimed in Paul (2000:103). At any rate, (20) cannot receive an interpretation such as
the one which may be allocated to (19), provided the latter is given an appropriate
context and a special intonation.
Indeed, an Instrument-Oblique reading along the lines sketched in section 3.1 is
simply impossible for (20), as that interpretation requires a circumstantial voice form
‘an…ana’ to surround a stem ‘sorat’. On the other hand, it is possible with the relevant
intonation to re-analyse the PP ‘amin’ity savony ity’ ‘with this soap’ in (19) as an
adjunct to the active voice verb manasa ‘to wash’: The PP will bear a rising intonation
while the rest of the sentence will carry another rising intonation, thereby giving rise to
a double rise peak. One crucial consequence of this special intonation pattern is that the
PP ‘Ø-amin’ity savony ity’ in (19) will be analysed as comprising a zero morpheme
symbolised by Ø and representing a non-past tense, i.e. in this case, the present tense.
The same type of double peak rise is totally impossible in (20). Furthermore, note the
presence of a third person subject ‘ny vehivavy’ ‘the women’ in (19) the active voice
verb ‘manasa’, which is ambiguous between a mere description of the event being
referred to and an exhortation to carry out an activity: It is well-known that an active
voice verb in Malagasy typically refers to the inception of the activity being described,
as reported in Randriamasimanana (1999:513), and as a result can mean something like
‘please start doing + the activity described by the verb’. Thus, with the exhortative
reading the PP ‘amin’ity savony ity’ in (19) can be construed as an adjunct to the verb,
hence the ‘nonpast’ gloss for the zero morpheme on ‘Ø-amin(a)’. Last, the third person
subject in (19) can then be re-construed as the addressee of a request for action given
the scenario depicted above – along lines sketched in Randriamasimanana (1985). Only
this interpretation is available to native speakers of Malagasy for sequence (19); the
non-contextual Instrument-Oblique reading indicated in the intended meaning is
definitely NOT possible.
80
Simple Sentences in Malagasy
3.3 Specifier-head relation & feature erasure
As shown in Randriamasimanana (2002:60), there is an interesting contrast
between the reading of a sequence comprising a passive voice and involving the
‘no…ina’ form (promotion of DO to Subject) as shown in (21), as opposed to the socalled circumstantial passive voice and involving the ‘an…ana’ or ‘i…ana’ form
(promotion of argument Oblique to Subject), as illustrated in (23).
(21) No-did(i)-in’
i Paoly ny
mofo.
7
Pass.past -stem.be.cut-suff.ina i Paoly Det bread
‘The (whole) bread was cut up by Paul.’
(22) N-andidy
t-amin’ ny
mofo Rabe.
Past-pref.an.root.cut perf-with Det bread Rabe
‘Rabe cut some of the bread.’
(23) N-an-didi-an-d-Rabe
ny
mofo.
Past-pref.an-stem.be.cut-suff.ana-ep.d-Rabe Det bread
‘Some of the bread was cut by Rabe.’
In (21) we have the ‘no…ina’ form of passive, whereas in (23), we have the so-called
circumstantial voice with ‘an…ana’. In both sequences, the subject ‘ny mofo’ is exactly
the same, but in (21) with ‘no…ina’, the relevant reading is one where the whole bread
was cut up while in (23) with ‘an…ana’, we obtain a partitive reading. This is the
outcome of specifier-head relation. In (22), the verb is in the active voice and we have
the same type of partitive reading encoded with the PP ‘t-amin’ny mofo.’
As suggested in Randriamasimanana (2004b), the phenomenon just illustrated is
related to verbal aspect, i.e. DURATIVE associated with the ‘an…ana’ or ‘i…ana’ form
as contrasted with PUNCTUAL aspect associated with the ‘no…ina’ form, for instance.
Precisely, this detail will help explain the ungrammaticality of the following types of
sequence, i.e. (24) found in Potsdam (2003, ex. 2c) and sentence (25) adapted from Paul
(2001, ex. 14).
(24) *Nividianan’i Bao ny akoho i Soa.
*N-i-vidian-an(a)
i
Bao ny akoho i
Soa.
Past-pref.i-stem.be.bought.for-suff.ana D.sg Bao the chicken D.sg Soa
Intended reading: ‘Soa was bought-for chicken by Bao.’
7
Judging from data presented in Randriamasimanana (2002:57-58), the presence of the passive
form of past tense ‘no’ or passive form of future tense ‘ho’ requires an overt subject within the
clause.
81
Charles Randriamasimanana
(25) *Nandrahoan’ ny lehilahy ny trondro ny vehivavy.
*N-an-(h)andraho-an(a)
ny lehilahy ny trondro
past-pref.an-stem.be.cooked.for-suff.ana the man
the fish
ny vehivavy
the woman
Intended reading: ‘The woman was cooked-for fish by the man.’
First note that in (25) the prefix an and the first syllable ‘an’ within the verbal stem
‘(h)andraho’ fuse together via haplology, but the ‘an…ana’ circumstantial passive form
is still quite apparent; in (24) the relevant form is ‘i…ana’ surrounding stem ‘vidian’.
Second, the definite article in front of both ‘chicken’ in (24) and ‘fish’ in (25) has been
discreetly dropped from the English translation. In fact, to make the sentences in (24)
and (25) perfectly grammatical, all we need to do is simply drop the definite article ‘ny’
in front of the DO ‘akoho’ ‘chiken’ in one and in front of DO ‘trondro’ ‘fish’, in the
other. The reason why this is crucial is straightforward: The physical presence of a nondefinite object will erase the DURATIVE feature associated with both verbal morphemes ‘an…ana’ and ‘i…ana’ when the head merges with its complement; this is very
important given the specifier-head relation outlined earlier. To see how this type of
feature erasure could be implemented, consult Randriamasimanana (2004b); for constructions very similar to the ones analysed above, see among others Sabel (2002),
Sabel (2003) and Potsdam (2005).
4. Location-Oblique
Thus, in Malagasy an Instrument-Oblique is an argument of the verb since the
latter necessarily comprises [+CONTROL] as part of its set of features. However the
situation with a Location-Oblique is much more complicated, as already described and
illustrated in Randriamasimanana (1999:515-516, ex. 8, 9, 10): First, a great deal
depends on whether the relevant PP is governed by a [+CONTROL] verb, in which case
it is an argument of the verb, whereas when the accompanying verb contains the
negative valued feature [-CONTROL], then the PP is an adjunct; second, verbal aspect
also seems to play a very important part in the process.
Randriamasimanana (1999:510, ex. 1) and (1999:515, ex. 8 & 9) shows that in a
declarative statement with an active voice verb like ‘go’, a PP denoting a ObliqueLocation is necessarily an argument of the verb which definitely contains the inherent
feature [+CONTROL] whereas the situation is more complex with intransitive verbs
like ‘mipetraka’ ‘to sit down’, which may have two totally different readings: On the
one hand, there may be some kind of Control involved if the event being described
82
Simple Sentences in Malagasy
means something like ‘to take the action of deliberately sitting down’ as opposed to the
other purely stative meaning, ‘to live, to stay’, in the sense of ‘to happen to be living
(somewhere), to happen to be staying (somewhere)’, which does not seem to involve
any Control whatsoever. In what follows, we consider questions involving a motion
verb, which one expects would be a rather characteristic [+CONTROL] type of verb
and contrast the situation of Wh-questions with that of Echo questions.
4.1 Wh-question
With Wh-questions involving a Location-Oblique in focus, a verb containing the
inherent feature [+CONTROL] must be in the ‘circumstantial’ passive voice. This
clearly contradicts one claim made in Sabel (2003:232, ex. 6a) as well as in Potsdam
(2004:246, ex. 5a). At any rate in the case of sequence (26), the a-passive form is the
relevant morpheme since we are dealing with a motion verb. What is absolutely crucial
here is that with this type of non-Echo question, there is only one intonation peak that
rises steadily from the beginning of the utterance to its very end. In addition, it is worth
noting that sequence (26) can appear discourse-initially.
(26) Taiza no nalehanareo?8
T-aiza
no
n-a-leha-nareo?
Perf-where focus
past-a.pass-root.go-by.you.plural
Literally: ‘Where is gone-to by you (all)?’
‘Where did you (all) go?’
In (26) the governing verb is in the a-passive voice and the morpheme t- is analysed as a
perfective aspect-marker; this indicates that this PP is an argument of the verb and that
the latter has to be in the passive voice: The verb mandeha ‘pres-go’ contains the
inherent feature [+CONTROL].
One interesting characteristic of a motion verb like the one in (26) is that it shows
that the feature [+CONTROL] can accommodate either DURATIVE or PUNCTUAL
8
Here is one non-Wh-question example of a ‘fronted’ Location-Oblique PP involving a nonmotion verb in the DURATIVE aspect & the circumstantial passive circumfix ‘an...ana’:
From http://www.taratramada.com/, 23 November 2005
Tany amin’ny Cap d’Antibes, Cote d’Azur no nanaovana ny fakana sary azy ireo.
T-any
amin’ny Cap d’Antibes, Cote d’Azur no
n-an-(t)aov-ana
perf-there at
Cap d’Antibes, Cote d’Azur focus past-an-stem.be.done-ana
ny
fakana sary azy ireo.
Nom. Taking picture
them
‘It was at Cap d’Antibes, Cote d’Azur where the taking of their pictures was done.’
83
Charles Randriamasimanana
aspect, a distinction already introduced in section 3.3. Indeed, on the one hand, the
combination of PUNCTUAL with [+CONTROL] will materialise as the passive
circumfix ‘no...ina’, as in (21); on the other hand, the combination of DURATIVE with
[+CONTROL] will yield the passive circumfix ‘an...ana’, as in (15) or ‘i...ana’, as in
(17). If we now add INCEPTIVE as a feature to the initial combination [+CONTROL]
& PUNCTUAL, then typically this latest combination will be encoded as a-passive, as
in (26), especially when reference is made to the INITIAL IMPULSE of whatever is
described by the verb. The above piece of Malagasy data suggests that a-passive
subsystem is midway between the ‘no…ina’ type (DO> Subject) and the ‘an…ana’ or
‘i…ana’ type (Oblique>Subject).
4.2 Echo question
In an Echo-type question involving a motion verb, as in (27) and (28) below, we
note a special intonation whereby the portion ‘taiza ianareo’ ‘t-aiza ianareo’ ‘past-where
you.plural’ in (27) bears an intonation rise and the rest, ‘no nandeha’ ‘focus past-go’,
has a second intonation rise as well. The same holds true of (28), where ‘taiza’ has an
intonation rise and so will ‘nandeha ianareo’. This practically means that as used here
the verb mandeha ‘pres-go’ contains the feature [-CONTROL].In addition, (27) and (28)
are appropriate as requests for confirmation, hence the label ‘Echo questions’ since
literally the speaker is quoting from a previous statement in the relevant adjacent pair of
utterances. This means an appropriate context9 is required.
(27) Taiza ianareo no nandeha?
T-aiza
ianareo no
n-an-(l>d)eha?
Past-where you.plur focus past-pref.an-root.go
‘Where did you say YOU went?’
9
84
Here is a similar, but authentic echo-type question from website http://www.serasera.com/
mpandefa (sender): Herimiafina 21-09-2005 21:42:20
Taiza ianao no nahita omby mifoka sigara, mirevy, na misotro cafe?
T-aiza
ianao no
n-ahita
omby m-ifoka
sigara, m-irevy,
past-where you focus past-root.see cows pres.root.smoke cigars, prs-rt.dream,
na m-isotro
cafe?
or prs-rt.drink coffee?
‘Where have you (ever) seen cows that smoke cigars, dream or drink coffee?’
Note the verb ‘n-ahita’ ‘past-root.see’ in the active voice with the ‘fronted’ ‘t-aiza’. This is
NOT a question at all, as the writer is certainly not expecting an answer. He is chiding his
interlocutor.
Simple Sentences in Malagasy
(28) Taiza no nandeha ianareo?
T-aiza
no
n-an-(l>d)eha
Past-where focus past-pref.an-root.go
‘Where did you say you went?’
ianareo?
you.plural
In both (27) and (28) the verb is in the active voice and the morpheme t- is analysed as a
past tense-marker and not as an aspect-marker; this signals that the PP ‘t-aiza’ ‘pastwhere’ is an adjunct to the verb. In general, neither is appropriate discourse-initially and
always presupposes a previous initial statement involving a location. Prototypically this
type of Echo-question features copying a portion of the relevant initial statement, which
usually corresponds to the verb; in other words, the verb ‘nandeha’ ‘past-go’ is a quote
from an earlier utterance within the discourse and that (27) and (28) are the second
utterances of two adjacent pairs.
5. Argument vs adjunct
As demonstrated in section 3.2 the notion of Control plays a crucial role in
determining the argument as opposed to the adjunct status of a given constituent, in
particular, where a PP is concerned. A PP governed by a typically transitive verb and
involving Control defined as some activity to be carried out intentionally and in a
deliberate manner is more likely to be an argument. By way of contrast, a PP
accompanying a STATIVE intransitive verb characterised by absence of Control, i.e.
not intentionally as shown in (33) section 5.2, is more likely to be an adjunct. In this
connection, note that in Randriamasimanana (1986:29-74), a whole section is indeed
devoted to this notion of Control, which brings with it a number of properties.
5.1 Argument PP, typically transitive V & control
In (29) we have a typical active voice sequence with an Instrument-Oblique PP,
tamin’ny antsy ‘with the knife’, whereas in (30) the verb has the circumstantial voice
‘an…ana’ morpheme along with the ‘fronted’ PP. By contrast, in both (31) adapted
from Paul (2000:38, ex. 26b) and (32) adapted from Pearson (2003, ex. 14a), the verb is
in the active voice and both sequences are irretrievably ungrammatical.
(29) Nandidy hena tamin’ny antsy i Bakoly.
N-an-didy
hena t-amin’
ny antsy
Past-pref.an-root.cut meat perf-with Det knife
‘Bakoly was cutting meat with the knife.’
i
Bakoly.
D.sg Bakoly
85
Charles Randriamasimanana
(30) Tamin’ny antsy no nandidian’i Bakoly hena.
i
Bakoly hena.
T-amin(a) ny antsy no
n-an-didi-an(a)10
Perf-with Det knife focus past-an-stem.be.cut-ana D.sg Bakoly meat
‘It was with the knife that meat was being cut by Bakoly.’
(31) *Tamin’ny antsy no nandidy hena i Bakoly.
T-amin(a) ny antsy no
n-an-didy
hena i
Bakoly.
Perf-with Det knife focus past-an-root.cut meat D.sg Bakoly
Intended reading: ‘It was with the knife that meat was cut by Bakoly.’
(32) *Amin’ny antsy no mamono ny akoho ny mpamboly.
Ø-amin(a)
ny antsy no
m-a(n>m)-(v)ono ny
akoho
Non-perf-with Det knife focus pres-an-root.kill
Det chicken
ny mpamboly.
Det farmer
Intended reading: ‘It is with the knife that the farmers kill the chickens.’
The reason for the ungrammaticaly of (31) and (32) is straightforward: The verb
nandidy ‘was cutting’ or ‘mamono’ ‘is killing’ describes a DELIBERATE kind of
ACTIVITY and as a direct result of this contains the feature [+CONTROL]; as a further
result, the PP ‘tamin’ny antsy’ is automatically an argument of the verb and the verb
must have the circumstantial voice morphology ‘an…ana’ surrounding the verbal stem
‘didi’ for the first and ‘vono’ for the second. Furthermore, as the action of cutting in (29)
is DURATIVE in nature, it is the ‘an…ana’ form of circumstantial passive encoding
DURATION, which is used.
5.2 Adjunct PP, typically intransitive V & absence of control
We now consider the case of sequence (33) involving a typical STATIVE
intransitive kind of verb, mirofotra ‘to explode’ with a non-fronted PP, adapted from
Rabenilaina (1985) and its counterpart (34) with a fronted PP, adapted from Randria-
10
86
In Paul (2000:27, ex. 12b), the claim is made that the a-passive typically promotes an
instrument to subject. However as noted in section 4.1 a-passive includes INCEPTIVE among
its features.
A-didy
ny
hena ny
antsy
a.pass-root.cut Det meat Det knife
‘The knife is used to cut the meat’.
It has to be noted that this sequence is only acceptable if what is being referred to here is an
electric knife. However the overwhelming majority of Malagasy speakers simply do not have
access to one.
Simple Sentences in Malagasy
masimanana (1998). It is certainly significant that this typical verb simply does not have
a passive voice form of any kind at all in the entire grammar of Malagasy.
(33) Nirofotra teo amin’ny tarehin’i Soa ny mony.
N-i-rofotra
t-eo
amin’ny tarehi-n’ i
Soa ny mony.
Past-pref.i-root.explode past-there on Det face-of D.sg Soa Det pimples
‘Pimples exploded on Soa’s face.’
(34) Teo amin’ny tarehin’i Soa no nirofotra ny mony.
T-eo
amin’ny tarehi-n’i
Soa no
n-i-rofotra
Past-there on the face-of D.sg Soa focus past-pref.i-root.explode
ny mony.
Det pimples
‘It was on Soa’s face that pimples exploded.’
The verb mirofotra ‘to explode’ does NOT describe a DELIBERATE kind of
ACTIVITY and is a STATIVE verb. It contains a negative feature [-CONTROL]; as a
result , in (33) the PP ‘t-eo amin’ny tarehi-n’i Soa’ ‘past-there on the face of Soa’ is an
adjunct: This is made obvious via the pattern ‘past-tense’ on the verb and ‘past-tense’
on the PP itself. When an adjunct is ‘fronted’, this does not require passive voice
morphology on the verb. In fact, in (34), the verb remains in the active voice; and
interestingly enough, no special intonation of the kind outlined earlier is necessary. For
further illustrations of the distinction between argument & adjunct in Malagasy, consult
Randriamasimanana (1999:509-528).
6. Focus phrase with ‘no’, fronting of argument vs fronting of adjunct
In addition to prototypical arguments, as discussed in section 5.1 and proto-typical
adjuncts, as outlined in section 5.2, there exist cases where one verb may contain either
feature [+CONTROL] or [-CONTROL] depending on their usage. Here as before, the
fact remains that the [+CONTROL] usage of a verb will force an accompanying PPLocation to be considered as an argument, whereas the inherent feature [-CONTROL]
usage of the verb will take the accompanying PP-Location only as an adjunct. Once
again, when an argument is ‘fronted’ circumstantial passive voice morphology must
appear on the verb, whereas when an adjunct is ‘fronted’, the verb remains in the active
voice.
87
Charles Randriamasimanana
6.1 Verbs with an optional feature [+CONTROL] or [-CONTROL]
The verb mianatra ‘pres-pref.i-anatra’ ‘to study’ in sequence (35) is a typical verb,
which may contain either inherent feature [+CONTROL] with a positive value or [CONTROL] with a negative value, depending on its usage. Depending on which value
of the feature is relevant, the sequence may mean one thing, i.e. the first reading
provided below or another, i.e. the second interpretation given.
(35) Mianatra ao an-davarangana i Soa.
Ø-ao
an-(l>d)avarangana i
Soa.
M-i-anatra
Pres-pref.i-root.study nonperf/nonpast-there at-balcony
D.sg Soa
Either ‘Soa is studying on the balcony’
or ‘(Please) Soa, go and study on the balcony.’
In fact, (35) is ambiguous between a [+CONTROL] reading and a [-CONTROL]
interpretation. In its [+CONTROL] reading, a somewhat elaborate context is required as
the sequence receives three different and somewhat rising intonation peaks from
beginning to end: Thus there is a first peak on head verb ‘mianatra’ ‘to study’, then
follows a second peak on the Location PP ‘ao an-davarangana’ ‘on the balcony’ and a
third peak on the subject ‘i Soa.’
By contrast, in its [-CONTROL] interpretation, the utterance has one steadily
falling intonation from beginning to end; we may take this to be the unmarked reading
for this verb, which after all is an intransitive verb. In other words, the [+CONTROL]
reading is something like ‘(Please) Soa, go and study on the balcony’, whereas the other
purely descriptive one, i.e. [-CONTROL] interpretation can be translated as ‘Soa is
studying on the balcony.’ In the [+CONTROL] reading, the Location PP ‘ao andavarangana’ ‘on the balcony’ is an argument of the verb and the zero morpheme Ø on
the preposition ‘Ø-ao’ is analysed as a non-perfective aspect-marker; by contrast, in its
[-CONTROL] interpretation, the Location-Oblique PP ‘ao an-davarangana’ ‘on the
balcony’ is an adjunct to the verb and the zero morpheme Ø on the preposition ‘Ø-ao’ is
analysed as a non-past tense-marker. Recall that the hallmark of an adjunction in
Malagasy is the parallel between two tense-markers, one on the verb and the other on
the accompanying PP.
6.2 Fronting of adjunct
Consider the [-CONTROL] reading of (35) first: the PP ‘Ø-ao an-davaran-gana’
‘on the balcony’ is an adjunct to the verb, the zero morpheme Ø being analysed as ‘non-
88
Simple Sentences in Malagasy
past’ tense and when it is ‘fronted’, the verb remains in the active voice, as shown in
(36).
(36) Ao an-davarangana no mianatra i Soa.
Ø-ao
an-davarangana no
m-i-anatra
i
Soa.
nonpast-there on-balcony
focus pres-prf.i-root.study D.sg Soa
‘It is on the balcony that Soa (usually) studies.’
Sequence (36) features the verb ‘mianatra’ ‘to study’ in its [-CONTROL] meaning, i.e.
a mere description and contrary to the situation in (37), certainly not an injunction nor a
request for action. Note the parallel between two tense-markers within the sentence, i.e.
‘Ø-ao’ ‘non-past tense-marker’ on PP and ‘present tense-marker’ on the verb ‘mianatra’
‘to study’, the hallmark of adjunction in Malagasy.
6.3 Fronting of argument
Now consider the [+CONTROL] interpretation of (35), which has been adapted
from Randriamasimanana (1998) repeated as (37) below: The PP ‘Ø-ao andavarangana’ ‘on the balcony’ is an argument of the verb, the zero morpheme Ø being
analysed as ‘non-perfective’ aspect-marker and when this PP is ‘fronted’, the verb must
take the circumstantial voice with the circumfix ‘i…ana’ surrounding the verbal stem
‘anar’, as shown in (38).
(37) Mianatra ao an-davarangana i Soa.
Ø-ao
an-davarangana i
Soa.
M-i-anatra
Pres-prf.i-root.study nonperf-there on-balcony
D.sg Soa
‘Soa, (go and) study on the balcony!’
(38) Ao an-davarangana no ianaran’i Soa.
Ø-ao
an-davarangana no
ianaran(a) i
Soa.
nonperf-there on-balcony
focus i-pass-stem.be.studied-ana D.sg Soa
‘Soa, (go) on the balcony and study there!’
Recall that given the special intonation described in section 6.1, sequence (37) is either
an injunction or a request. The verb ‘mianatra’ ‘to study’ as used here contains the
inherent feature [+CONTROL], which automatically makes the accompanying PP an
argument.
89
Charles Randriamasimanana
7. Incorporation
One prototypical adjunct within a Malagasy sentence is Time-Oblique. But even
this kind of Oblique adjunct can optionally be incorporated, i.e. ‘turned’ into a proper
argument first and thus made part of the initial core of the clause. This process seems to
be related to verbal aspect and crucially depends on the compatibility of the set of
features accompanying a head verb with [+CONTROL]. This process is missing in
considerations of typical adjuncts in Paul (2000:91-130), who claims that an ‘adjunct
can be promoted directly to subject’ and this is somehow echoed in Pearson (2001:141149). While such a claim may be able to accommodate data such as (41) below with a
circumstantial voice passive ‘i..ana’ form, it would be rather difficult to explain why
certain verbs such as mirofotra ‘to explode’ in (33) and (34) above simply do not have
any passive form of any kind at all.
7.1 Initial adjunct PP
A prototypical Malagasy adjunct is Time-Oblique, such as the PP ‘t-amin’ny
herin’ny sabotsy lasa’ ‘past-on the return of the Saturday before last’ ‘the Saturday
before last’ found in (39).
(39) Niala teto Rabe tamin’ny herin’ny sabotsy lasa.
N-i-ala
t-eto
Rabe t-amin’ny
Past-pref.i-root.leave perf-here
past-on the
sabotsy
lasa.
Saturday
gone
‘The Saturday before last, Rabe left here.’
herin(a)’
return-of
ny
Det
One interesting feature of a typical adjunct PP is that it is usually located to the right of
the grammatical subject, in the case of (39) ‘Rabe’ and is part of a steadily falling
intonation pattern from the beginning to the end of the utterance, with only a slight
pause in front of ‘tamin(a): Thus in (39), Rabe is the subject and the relevant PP ‘tamin’ny herin’ny sabotsy lasa’ ‘the Saturday before last’ shows up after the subject
‘Rabe’; also note that the morpheme t- on the preposition ‘amin(a)’ is a past tensemarker, not a perfective aspect-marker.
90
Simple Sentences in Malagasy
7.2 Intermediate structure
The sequence in (39) is the unmarked version of the utterance, while (40) is its
marked counterpart, as it involves a first big pause after the Time-Oblique ‘teto’ ‘perfhere’ and a second big pause after the verb ‘lasa’ ‘gone’.
(40) Niala teto tamin’ny herin’ny sabotsy lasa Rabe.
N-i-ala
t-eto
t-amin(a) ny
Past-pref.i-root.leave perf-here past-on
the
lasa
Rabe.
gone Rabe
‘Rabe left here the Saturday before last.’
herin(a) ny sabotsy
return-of. Det Saturday
As (40) shows, it is possible to have the Time-Oblique PP ‘t-amin’ny herin’ny sabotsy
lasa’ ‘the Saturday before last’ immediately to the left of the grammatical subject
‘Rabe’. This now makes such a displaced PP eligible for incorporation into the adjacent
VP headed by the verb ‘niala’ ‘past-leave’, once the ‘past-tense’ feature for morpheme
‘t-’ has been switched to ‘perfective aspect’ feature; this is a plausible move since
morpheme ‘t-’ is ambiguous between those two values just referred to. For a
justification of the necessity of adjacency in Malagasy, consult Randriamasimanana
(2004a).
7.3 Fronting after incorporation
One crucial detail apparently associated with incorporation of a Time-Oblique is
that such a process is invariably accompanied by ‘fronting’ of the incorporated PP, in
this case, the Time-Oblique ‘tamin’ny herin’ny sabotsy lasa’ ‘the Saturday before last’,
as shown in (41). A non-fronted PP in (41) would make it ungrammatical.
(41) Tamin’ny herin’ny sabotsy lasa no nialan-dRabe teto.
T-amin(a)’ ny herin(a)’ ny sabotsy
lasa
no
Perf-on
Det return-of the Saturday gone focus
n- ial(a)an(a)-dRabe
t-eto.
past-pref.i-stem.be.left-ana-epenthetic.d-Rabe
perf-here
‘It was on the Saturday before last that Rabe was leaving here.’
In (41) the PP ‘t-amin’ny herin’ny sabotsy’ is a derived argument of the verb and that
now DURATIVE circumstantial voice morphology indicated by the circumfix ‘i…ana’
91
Charles Randriamasimanana
is obligatory around the verbal stem ‘ala’. With this example, the verb ‘nialana’ is in the
circumstantial passive and we have a case of so-called internal viewing perspective and
the verb contains the feature [+CONTROL].
7.4 Fronting without incorporation
On the other hand, in sequence (42) the PP ‘tamin’ny herin’ny sabotsy’ is an
adjunct to the verb and as a consequence, ‘fronting’ it does not require passive morphology at all. And, of course, the adjunct can be left non-fronted, as in (39).
(42) Tamin’ny herin’ny sabotsy lasa no niala teto Rabe.
T-amin(a) ny herin(a) ny sabotsy
lasa no
n-i-ala
Past-on
the return-of the Saturday gone focus past-i-root.leave
t-eto
Rabe.
perf-here Rabe
‘It was on the Saturday before last that Rabe left.’
Here the verb niala ‘past-leave’ ‘left.PUNCTUAL aspect’, contains the negatively
valued feature [-CONTROL] and we have a case of the so-called external viewing
perspective indicated in English by the use of the non-progressive aspect. For further
details on Malagasy incorporation, consult Randriamasimanana (2004a:421-422).
8. Conclusions
In this paper, it has been shown that simple Malagasy sentences may typically
comprise a core as well as a periphery. Thus core will comprise a DO in the case of a
transitive verb as in (1) and an IO in that of a di-transitive verb as in (7) at least, in an
active voice sequence whereas periphery will include typical adjuncts such as TimeOblique, for instance, as shown in (39). In-between those two quite distinct and welldefined zones, we have a rather fuzzy area where optionality of a given constituent is
possible but where an Instrument-Oblique as in (29) always behaves like an argument
as such a PP is typically constituent-selected by a [+CONTROL] verb. Likewise, with a
Location-Oblique, pretty much depends on the type of verb involved and whether the
latter contains the positively valued feature [+CONTROL] or not: When a given PP is
selected by a [+CONTROL] predicate, the PP is an argument, as in (26); whereas if the
PP is accompanied by a negatively valued [-CONTROL] predicate, the PP is only an
adjunct, as in (27) or (28). This confirms results already reported in Randriamasimanana (1999) with respect to motion verbs and Location-Oblique PPs in declarative
92
Simple Sentences in Malagasy
statements. However certain verbs like mianatra ‘to study’ in (35) may have a positively
valued feature or a negatively valued feature depending on their wider context of use, as
reflected in the intonation pattern of the entire utterance.
At any rate, even though it is not part of the core but rather inside the so-called
fuzzy area alluded to in section 1, an Instrument-Oblique is a ‘derived’ argument of its
governing verb, which necessarily contains the feature [+CONTROL]. In fact, neglect
of this feature has led to consequences described in section 3.2. The situation is slightly
more complex with a Location-Oblique: This type of PP is not part of the core either
and is also located inside the same fuzzy area; however, in this case, here either
[+CONTROL] with a motion verb, as shown in section 4.1 is plausible just like [CONTROL] with a STATIVE verb, as illustrated in section 5.2. This means that a
Location-Oblique can be an argument with a [+CONTROL] verb, as in (26) but an
adjunct with a negatively valued feature [-CONTROL] verb, as in (33). Above and
beyond this, it is a fact that the passive voice system in Malagasy presents a tripartite
division into a ‘no...ina’ passive, on the one hand and an ‘an/i...ana’ passive on the other;
between those two subareas lies another ‘fuzzy’ zone, where the quintessential
characteristic feature of a-passive manifests itself.
As outlined in section 3.3 verbal aspect does play a major role in the grammar of
this language and this is most evident in the passive voice system. Thus it seems that the
feature [+/-CONTROL] inside a verb determines to a large extent whether an accompanying PP is likely to be an argument, as in (12) or an adjunct, as in (33), whether the
verb is in the active or the passive voice. Furthermore if one additional feature, i.e.
INCEPTIVE, a significant characteristic of a-passive, is added to either PUNCTUAL or
DURATIVE, as suggested relative to (26), then it becomes possible to switch from one
subsystem of passive, i.e. ‘(no)...ina’ form, as in (21), involving promotion of a DO to
Subject to the other subsystem, i.e. ‘i/an...ana’ form, as in (23), involving promotion of
an Oblique to Subject.
Somehow linked to quantification and therefore related to the contrast between
PUNCTUAL and DURATIVE verbal aspects is the selection between the two values of
Malagasy ‘ny’ -- assuming a type of specifier-head relation between verbal inflections
and the grammatical subject as envisaged in section 3.3: In its Det value, ‘ny’ may refer
to all members of a given set, as in (13) -- where it just so happens that here the set only
comprises one member -- whereas in its Topic value, ‘ny’ designates only some
members NOT all, of a given set, as illustrated in (3), (4) and (6). This will go some
way towards explaining why in Malagasy it is possible to front a non-definite DO even
though the verb remains in the active voice, a fact which is being denied in the
published literature so far, as discussed in section 2.1.1.
93
Charles Randriamasimanana
In addition, it is also worth noting that even typical adjuncts like Time-Oblique
may find their set of inherent features compatible with [+CONTROL], and resort to
incorporation, thereby undergoing a switch. Nevertheless it must be emphasised that
this possibility rests with the head verb and that not all verbs can undergo this process;
and precisely this is where the feature [+CONTROL] makes a crucial difference, as
apparently a prototypically STATIVE verb like mirofotra ‘present-explode’ is fundamentally incompatible with this feature.
Last but not least, it is to be hoped that a number of issues raised in this paper,
such as the correlation between the [+CONTROL] feature and wh-questions will help
shed some light on the Malagasy data proposed in numerous papers on this language, to
include Potsdam (2004), Potsdam (2005), Sabel (2002) and Sabel (2003). However an
in-depth study of the intricacies involved in the myriad aspects of all the phenomena
evoked here lies beyond the scope of a short paper like this one and will have to await
further research.
94
Simple Sentences in Malagasy
References
Keenan, Edward L.1976. Remarkable subjects in Malagasy. Subject and Topic, ed. by
Charles N. Li, 303-334. New York: Academic Press.
Law, Paul. 2005. Questions and clefts in Malagasy. UCLA Working Papers in
Linguistics 12: Proceedings of AFLA-12, ed. by Jeffrey Heinz and Dimitris
Ntelitheos, 195-209. Los Angeles: UCLA.
Paul, Ileana, and Eric Potsdam. 2004. How to sluice in the wh-in-situ language
Malagasy. Paper presented at the 40th Annual Meeting of the Chicago Linguistic
Society. Chicago: Chicago Linguistic Society.
Paul, Ileana. 2000. Malagasy Clause Structure. Montréal: McGill University dissertation.
Paul, Ileana. 2001. Concealed pseudo-clefts. Lingua 111.10:707-727.
Paul, Ileana. 2002. An explanation of extraction asymmetries in Malagasy. Linguistic
Variation Yearbook 2.1:99-122. Amsterdam: John Benjamins.
Pearson, Matthew. 2001. The Clause Structure of Malagasy: A Minimalist Approach.
Los Angeles: UCLA dissertation.
Pearson, Matthew. 2003. Malagasy voice morphology and agreement. AFLA-10
handout. Honolulu: University of Hawai‘i at Mānoa.
Potsdam, Eric. 2003. Ellipsis identity and Malagasy sluicing. Paper presented at AFLA10. Honolulu: University of Hawai‘i at Mānoa.
Potsdam, Eric. 2004. Wh-questions in Malagasy. Proceedings of AFLA-11, ed. by Paul
Law. Berlin: ZAS.
Potsdam, Eric. 2005. The clausal typing hypothesis and optional wh-movement in
Malagasy. Colloque de Syntaxe et Sémantique handout. Paris: Université Paris VI.
Rabenilaina, Roger-Bruno. 1985. Lexique-Grammaire du Malgache. Paris: Université
Paris VII dissertation.
Rajaona, Siméon. 1969. Takelaka Notsongaina, Vol. 2: Lahatsoratra Tsotra. Fianarantsoa:
Ambozontany.
Randriamasimanana, Charles. 1985. Tense/aspect and locatability in Malagasy. University
of Melbourne Working Papers in Linguistics 11, 109-136. Melbourne: University
of Melbourne.
Randriamasimanana, Charles. 1986. The Causatives of Malagasy. Oceanic Linguistics
Special Publication No. 21. Honolulu: University of Hawai‘i Press.
Randriamasimanana, Charles. 1994. Malagasy and universal grammar. Proceedings of
the Seventh International Conference on Austronesian Linguistics, ed. by Cecilia
Odé and Wim Stokhof, 483-496. Leiden: Leiden University.
Randriamasimanana, Charles. 1998. Passive as incorporation in Malayo-Polynesian
Languages. Invited talk to AFLA-5. Honolulu: University of Hawai‘i at Mānoa.
95
Charles Randriamasimanana
Randriamasimanana, Charles. 1999. Clausal architecture and movement verbs in
Malagasy. Selected Papers from the Eighth International Conference on Austronesian
Linguistics, ed. by Elizabeth Zeitoun and Paul Jen-kuei Li, 509-527. Taipei:
Institute of Linguistics, Academia Sinica.
Randriamasimanana, Charles. 2002. Binary branching and null subjects in Malagasy.
Language and Linguistics 3.1:43-78.
Randriamasimanana, Charles. 2004a. Malagasy and Formosan languages: a comparison.
Language and Linguistics 5.2:409-431.
Randriamasimanana, Charles. 2004b. Relation tête-spécifieur et analyses en traits
pertinents en malgache. Faits de Langues — Les langues austronésiennes, No. 2324, ed. by Elizabeth Zeitoun, 271-280. Gap: Ophrys.
Sabel, Joachim. 2002. Wh-questions and extraction asymmetries in Malagasy. MIT
Working Papers in Linguistics 44: Proceedings of AFLA-8, ed. by A. Rackowski
and N. Richards, 309-323. Cambridge: MITWPL.
Sabel, Joachim. 2003. Malagasy as an optional multiple wh-fronting language. Multiple
Wh-fronting, ed. by Cedric Boeckx and Kleanthes Grohman, 229-254. Amsterdam:
John Benjamins.
http://www.haisoratra.org/ Ny Haisoratra Malagasy on-line (Malagasy Arts & Letters
on-line), Thursday 6 March 2003, put on-line by Hajanolalaina.
http://www.midi-madagasikara.mg/ Midi Madagascar, dated 21 October 2005.
http://www.serasera.com/ 21-09-2005 21:42:20.
http://www.taratramada.com/ 23 November 2005.
http://www.wanadoo.mg/forums/read.php? Re: Voyage en Inde, Par: lalaiko, Date: 28
Septembre 2005 - 10:58.
96
LANGUAGE AND LINGUISTICS
LANGUAGE AND LINGUISTICS is a department of the INSTITUTE OF LINGUISTICS,
ACADEMIA SINICA. It furthers the Institute’s objective of excellence in research, scholarship, and
education by publishing pioneering articles and monographs in Linguistics and other related fields. All
publications have been reviewed rigorously according to academic standards.
Copyright © 2006 by INSTITUTE OF LINGUISTICS, ACADEMIA SINICA
128, Sec. 2, Academia Road, Nankang, Taipei 11529, Taiwan, R. O. C.
http://www.ling.sinica.edu.tw
All rights reserved.
LANGUAGE AND LINGUISTICS MONOGRAPH SERIES NUMBER W-5
Henry Y. Chang, Lillian M. Huang, and Dah-an Ho (Editors):
STREAMS CONVERGING INTO AN OCEAN:
FESTSCHRIFT IN HONOR OF PROFESSOR PAUL JEN-KUEI LI
ON HIS 70TH BIRTHDAY
《語言暨語言學》
《語言暨語言學》隸屬於中央研究院語言學研究所，以出版語言學及相關領域之最新
研究成果為宗旨。所有出版品均經過嚴格學術審查。
《語言暨語言學》專刊外編之五
百川匯海：李壬癸先生七秩壽慶論文集
編輯：張永利
黃美金
何大安
出版及發行：中央研究院 語言學研究所
115 台北市 南港區 研究院路 2 段 128 號
http://www.ling.sinica.edu.tw
版權所有 翻印必究
印刷：文盛彩藝事業有限公司
初版：中華民國九十五年十月 [2006]
定價：新台幣900元／US$50
ISBN-13: 978-986-00-6784-2 (精裝)
ISBN-10: 986-00-6784-8 (精裝)
GPN: 1009502736